One of my major research projects has been to explore the potential for improving weather prediction using the pressure sensors in smartphones, something I have talked about in previous blogs.

A number of Android phones has decent barometer sensor in them, including the highly popular Samsung Galaxy series. A Windows-based phone, the Nokia 1020, also has a pressure sensor. Amazingly, tens of millions of phones in North America have pressure sensors!

But we now have a major new addition to the pressure-enabled smartphone stable: the new iPhone 6.

Both iPhone 6 models have a pressure sensor

According to online reports, the iPhone6 has a Bosch BMP280 sensor (see image), with fairly good numbers: absolute accuracy of +-1 hPa and relative accuracy for pressure changes of +-.1 hPa (normal sea level pressure is roughly 1013 hPa). To give you a better idea of the accuracy of this barometer, the average decrease in pressure with height near sea level is 1 hPa per 8 meters (26 ft) .

Bosch Pressure Sensor

So why am I so excited by smartphone atmospheric pressure sensors? Why do I believe they have revolutionary potential for weather prediction?

Because they offer the chance to get a extraordinary density of pressure observations, which provides the potential to describe small scale atmospheric structures. Structures we need to knwo about if we are to predict key weather features like strong thunderstorms.

Let me illustrate how many surface pressure observations there are. Currently, I am getting real time feeds from two small companies, Cumulonimbus, Inc, (with the Pressurenet app) and OpenSignal (with their WeatherSignal app). Right now, there are about 115,000 pressure observations coming in per hour from these innovative firms, (90% of them coming from the PressureNet app). Here is the map.

Amazing coverage from DC to Boston, and around other major cities. But there is substantial density beyond, particular east of the Mississippi and along the West Coast.

But keep in mind that this is less than one-hundredths of the smartphones with pressure sensors. Yes, 1/100. In a few years there will be at least 100 millions smartphones over North America with pressure sensors. So the additional of the iPhone 6 pressure sensors will only accelerate the growth, with Cumulonimbus, Inc already working on an appropriate iPhone pressure app.

If we could collect, say pressure from 1/10 of the phones with barometers, the eastern U.S. would be virtually covered and only few uninhabited western areas would have pressure observations.

So why would these pressure sensors be a boon for weather prediction? Because the numerical weather prediction is now going to smaller and smaller scales, and meteorologist are trying to do much better in predicting what will happen during the next few hours (called Nowcasting).

To forecast fine-scale weather features (like thunderstorms), you need a fine-scale description of the atmosphere, and the current observational network is often insufficient. We need millions of observations per hour over the U.S. to do the job. Same situation in China, Europe, and the rest of the world


And pressure is the perfect surface observation: it reflects the deep structure of the atmosphere and has less exposure problems than temperature or wind. Pressure can be measured inside our outside a building, in your pocket or hanging on your belt. A number of number experiments have shown that surface pressure measurements alone can produce a very good THREE-DIMENSIONAL description of the atmosphere. Almost sounds like magic.

I believe that dense pressure observations could radically improve weather prediction, and early numerical experiment support this claim.

But the big promise will NOT be met until we find a way to collect a higher percentage of the smartphone pressures.

Google could obviously do this. I have approached Google about capturing pressure observations on Android phones, but Google management does not seem interested (but a number of Google engineers have been very supportive).

Another approach would be for Samsung or Apple to preinstall the code for capturing and transmitting the pressure information

Or a group with a very popular app (like the WeatherChannel or the Weatherunderground) could include the relevant code .

Anyway, it is frustrating...a huge improvement in weather prediction is possible. Their is no major technical hurdle. The pressure sensors are in place. And we have not put it together. Maybe soon....